In any case, these are dreams that stand out to me, and the visuals are often the most interesting bit. Everything appears crystal clear and detailed as I dream, and after getting up, the imagery stays etched in my mind for a while.

In some cases, I’ve documented these rare dreams with a few sentences. To my surprise, it turned out that just a short summary is enough to restore the imagery in my mind later – in some cases even years later. After waiting long enough, the impression resembles any other visual memory, except that it feels a bit more alien.

I’ve sometimes considered making sketches of these dreams, but it always seemed too cumbersome. Could AI help me out here? It’s been a while since I took image generation models on a test run, so I decided to go for another tour. This time I wanted to see if AI would give me a decent approximation of dream visuals without spending a lot of time on prompt engineering.

These are the results.

Dream about an adventure

Prompt

A monstrously large spider monster in a cave, preying on a metallic, quicksilver-like, weirdly shaped blob. The blob is stuck to the ceiling some way above. A friendly person looks in through the door crack of a thick, old, wooden door. The door is set in the cave wall.

Result

Background

This dream was probably inspired by a re-read of Lord of the Rings. (At a later point in my life, it might also have been dubbed “a visit at the in-laws”.) I tried different prompts to get the metallic blob to look ever so slightly anthropomorphic, just as it was in my dream, but such instructions would always mess up other parts of the image.

Dream about feeling crushed

Prompt

An Asian woman sitting on a plain green bench without a backrest. She is wearing rings with very large and colorful gemstones on her hands. Opposite of her, lying flatly on his back on another similar bench, a man is being crushed by a black box. The man holds a cornet with strawberry ice cream in his dangling hand.

Result

Background

This dream was inspired by a girl who preferred another guy over me. They didn’t actually end up exchanging rings and instead split up after a year or so. Ha-Ha!

Dream about fucking up at school

Prompt

A pupil in the hallway of a school, surrounded by a few other pupils. His face shows that he has just realized there is a math exam today and he hasn’t prepared for it.

Result

Background

I used to have this or similar dreams for years after I finished school. In many of these cases, I “realized” during the dream that I had missed something important at school, and thus my college diplomas would also be invalidated! I’d have to go back to school to re-do the final school years, and somehow also juggle this task with my job and family life. Hence the facial expression in the image is hardly exaggerated …

Dream about an old man

Prompt

An old man in a grey and brown checkered suit, feeding ducks with popcorn from a trash bag.

Result

Background

Admittedly, this dream isn’t very outlandish. Some whacko giving inappropriate food to a random type of bird? You know, if you’ve lived in Berlin for long enough, a scene like this may almost give you a fuzzy and homey feeling.

Dream about dancing in the sky

Prompt

A man in black clothes flying through the evening sky and dancing ecstatically at the same time, even contorting himself. The sound of airflow is like a synthesizer pad. The landscape far below is passing by so rapidly that it’s a bit blurry.

Result

Background

I had a high fever when I dreamt this. The dream stood out for being quite synaesthetic and also including some very specific music, full of driving rhythms and suspenseful melodies. I tried to recreate that music in Fruity Loops the next morning, but the sounds evaporated from my memory more quickly than I could rig them up. If I ever have such a musical dream again, I’ll try using a music generator like Suno for a quick shot.

Dream about Crete

Prompt

A long barren coastline of Crete, with steep cliffs and several white-washed, high-rising monasteries and several very high arches dispersed across it.

Result

Background

I’ve visited Crete a few times as a kid. There were more and taller arches in the dream, like in a photo-realistic version of a de Chirico painting. However, any changes I made to the prompt made other bits worse, e.g. the coastline would turn less Cretan, so I left it at this.

Dream about misery

Prompt

The stairwell of a rundown prefabricated building. It is almost too dark to see anything. On an intermediate storey, there is an entrance to an elevator. The door is missing. Inside the elevator, there is a piece of burnt tinfoil on the ground, as well as a black Ghettoblaster. Both are just about visible in the dark. In the back of the elevator, there is a subtle shadow of a cowering dark figure.

Result

Background

I was probably processing my regular encounters with drug addicts on Berlin’s streets in this one. (And while this dream had a nightmarish touch, it’s worth noting that the increase in foil smoking in recent years is arguably a good trend, as it is a lot less harmful to the addicts than syringes.)

Dream about Darth Vader

Prompt

Darth Vader inhaling gummy bears through the grille of his helmet, profile view.

Result

Background

This dream was unusual in that I honestly had to chuckle when I woke up. There were gummy bears at the party the night before. Not sure where that Vader guy came from. Probably just existential dread or something.

Conclusion

Do some of these images look like what I saw in my dreams? Yes – somewhat! A few come quite close, which made it ever the more frustrating that I could never get the final 20% right. As you may have guessed, all these prompts went through a few iterations, and I tried each of them in two or three models. Yet in all of the cases, I hit a brick wall at some point.

I’m not saying this is the model’s fault. Well, okay, in some cases it is. E.g., for the dream with the two green benches, I couldn’t get the benches to stand opposite of each other and still preserve other aspects of the image, no matter which way I described it, and this doesn’t sound like a very challenging requirement at face value. It would have been a lot easier to explain this to a human artist. I’d probably just use my hands to demonstrate the spatial arrangement.

I could certainly still get better at prompting, but in some cases, there might be a limit to language, no matter whether the recipient is a human or a machine. Just because a highly specific image exists in someone’s head, that doesn’t mean there is an exact and unambiguous way to get it out of there in one go, without iterating on intermediate results. We’re not pixel enumerators, after all, and I guess that’s exactly why we can have such rich, multi-layered, absurd, yet often pretty experiences.

If I wanted these images to really approximate the “originals”, I’d either have to whip out photoshop and at the very least stitch together multiple generated images, or iterate and rely heavily on in- and out-painting. So at this point, I’m not sure using AI beats sketching or writing a short summary for the purpose of keeping a hold of dream visuals. Sketching probably still has a better balance of speed and authenticity, though of course generated images look nicer. Another thing worth trying might be to first make a sketch, and then augment it with AI. Or, you know, just reading out the images directly from the brain.

The models will certainly struggle to interpret song lyrics poetically, so it will be interesting to see whether the output will reflect the mood of the songs at all.

It’s also interesting to see how well each individual model deals with varying levels of metaphorical language and non-standard prompts.

To find out, I’ve taken ten very different pieces of music, fed their lyrics into Stable Diffusion, Midjourney, and Dall-E 3, and compared the results.

To keep things fair, I entered only the same snippets of lyrics as input into each of the models, without embedding them into any other prompt, and always used the initial results.

The Beatles - Lucy in the Sky with Diamonds

I suspected that well-known songs might produce different results compared to little-known songs because their lyrics might appear in the training data, either as image titles or as OCR-processed texts in the images themselves. Hence I started with the best-selling band of all. Lucy in the Sky with Diamonds is a song about an LSD trip, as the title gives away without too much subtlety. The lyrics are heavy on the visual, so I expected the results to capture the mood of the song quite easily.

Prompt

Picture yourself in a boat on a river
With tangerine trees and marmalade skies
Somebody calls you, you answer quite slowly
A girl with kaleidoscope eyes

Results

Stable Diffusion

The boat and the river are there, as are the dominant colors, but it somehow reminds me of a dull boat trip that someone’s been forced to do with their grandma.

Midjourney

With its vibrant colors and symbolic flair, this looks a lot more psychedelic and is more evocative of an intense experience.

Dall-E

Dall-E tried to incorporate the lyrics into the image and was the only one to show the face of the girl. The girl wears a lot of vaguely kaleidoscopic stuff, but her eyes are completely normal. I found it interesting that none of the models picked up on the most visually striking and distinctive detail in the lyrics – the kaleidoscope eyes.

Nirvana - Smells Like Teen Spirit

According to Cobain himself, the lyrics for Smells Like Teen Spirit are “kind of meaningless, because the point is that rock’n’roll is supposed to be meaningless”. Still, you’d expect a good visualization to evoke some kind of teenage angst or boredom.

Prompt

With the lights out, it’s less dangerous
Here we are now, entertain us
I feel stupid, and contagious
Here we are now, entertain us

Results

Stable Diffusion

A young guy with a grungy style and a disillusioned facial expression. Spot on.

Midjourney

What is this? Harry Potter and the Cave of Dimmed Lamps? Well, it has a young guy and the mood is dark, so it’s not completely off point.

Dall-E

This looks like a bunch of zombies going to a rock concert. The inset lyrics go well with that, because if you try to read them out loud you’ll feel approximately as “stupid and contagious” as a zombie…

Metallica - King Nothing

King Nothing is a song about hubris and empty pursuits. The lyrics are more metaphorical than in the previous examples, so it will be interesting to see how the models deal with that.

Prompt

And it all crashes down
And you break your crown
And you point your finger, but there’s no one around
Just want one thing, just to play the king
But the castle crumbled and you’re left with just a name

Results

Stable Diffusion

This forlorn figure in front of a derelict castle has more of a wistful air. The blame apportioned in the lyrics is not really reflected in the image.

Midjourney

This one I found really interesting. It doesn’t visualize a single noun from the prompt, but seems to find a nice symbolism for the song anyway: overplaying your cards. Just a coincidence?

Dall-E

I thought this one was pretty good. Perhaps the king remains a bit too self-contented to really fit the lyrics, but rendering the castle at the size of a plaything is a nice “idea” that suits the theme.

Gloria Estefan - Conga

Conga is in many ways an archetypical dance track. The lyrics don’t play a major role here. They mostly try not to get in the way of the rhythm and mood.

Prompt

It’s the rhythm of the island
And like sugarcane so sweet
If you want to do the conga
You’ve got to listen to the beat

Results

Stable Diffusion

The Stable Diffusion images often take on this “old master” style. This one reminds me a bit of Gauguin or Diego Rivera. There’s plenty of dynamism, but I’m not sure I feel invited to this particular party.

Midjourney

As in the previous image, none of the musical instruments really resemble a conga. But don’t you just wanna hug this guy? And join a conga line with him?

Dall-E

Is that sugar cane wrapped in bacon? It’s an adorable tiny party nevertheless.

Kanye West - Power

One of Kanye’s biggest hits, Power is actually a bit of an odd song, blending maximum braggadocio with musings on inner struggles. For the prompt I went with the more boastful part.

Prompt

I’m living in that 21st century
Doing something mean to it
Do it better than anybody you ever seen do it
Screams from the haters, got a nice ring to it
I guess every superhero need his theme music

Results

Stable Diffusion

Stable Diffusion picked up on the word “superhero”. Apparently it couldn’t decide which one it preferred.

Midjourney

This one resembles a movie poster. I like the chunky headphones. The hero might be using those to listen to the screams of his haters on repeat.

Dall-E

I’ll let this one stand for itself. Let me just say, I’m pretty sure this is not the type of “power” Kanye had in mind…

Kinski - Newport

Let’s try something more artsy where the intention is not so obvious. If anything, I would say that Kinski’s Newport has a vaguely intense and breathless ambience. It’s also a pretty unknown band and song, so let’s see if that has any effect on the outcome.

Prompt

Run
to the beach-house
stay there
’til you get my phone-call

Results

Stable Diffusion

Yep, she’s running to the beach-house, but will she wait there until she gets that ominous phone call? Hard to tell. The chilled-out dog kinda ruins the tension either way.

Midjourney

This is the type of small beach-house that I had in mind. The only thing is that with so many neighbors nearby, waiting there doesn’t feel as exciting as it should.

Dall-E

Finally! A Dall-E production that is not cluttered with gibberish text! The beach-house is more of a mansion than I would expect, but maybe it’s a haunted mansion? The guy even has a phone for that upcoming call. But why is he hanging onto his business class luggage amidst all that drama? Are there perhaps some nuclear codes in that suitcase? Consider me intrigued.

Joanna Newsom - Go Long

Joanna Newsom’s music often deals with complex emotions, and her rich imagery can be tricky to unpick. What will the models make of Go Long?

Prompt

There’s a man
who only will speak in code,
backing slowly, slowly down the road.
May he master everything
that such men may know
about loving, and then letting go.

Results

Stable Diffusion

This is a very literal interpretation, but the mood is sombre and thoughtful enough.

Midjourney

In what might be a coincidence, this one includes antique clothing and an autumnal scenery that resemble those found in some of Newsom’s promotional images. If you look closely, you can see the lights of a cozy home in the background fog, which goes well with the song’s theme of painful separation.

Dall-E

This might be a bathroom tile that has been put through a typewriter.

Clouddead - Physics of a Bicycle

Now for another little-known and challenging bit. Clouddead are a hip-hop group and known for their abstract lyrics. It’s not obvious what their song Physics of a Bicycle is about. My take is that it tries to capture the feeling of amazement that can occur while being high.

Prompt

Mother Nature made the airplane
And submarine sandwich
With the steady hand and dead eye of a remarkable sculptor
She shed her mountain-turning training wheels
For the convenience
Of a moving sidewalk

Results

Stable Diffusion

I can’t see much in this image except a distorted plane. The other elements are missing and there is not much of a symbolic touch. With a lot of benevolence you could call the result enigmatic.

Midjourney

I didn’t think you could merge a plane, a submarine, and a cornucopian sandwich into one, but here we are. It’s just a bit over the top compared to the more low-level psychedelic experience that the song hints at.

Dall-E

This one is my favorite in this round. It’s like mother nature actually made the airplane as a gift and even put her signature on it in almost correct English.

Electric Six - I’m the Bomb

Let’s try something that is a bit ironic and over the top. Electric Six often play with exaggerated machismo, and I’m the Bomb is no exception. Will the models pick up on the tongue-in-cheek tone?

Prompt

Now who elected you judge and jury
In the body of a beautiful girl?
And I suspect heavy gerrymandering
At the single’s bar
And now, I’m a man with conversations skills
And I’m a man with hundred dollar bills

Results

Stable Diffusion

A man with dollar bills, that’s about it.

Midjourney

This one has done better to capture the machismo and the beauty it longs for. The text looks completely random but doesn’t detract too much.

Dall-E

Dall-E somehow managed to include all the things mentioned in the prompt, yet the result is kinda boring.

The Beards - If Your Dad Doesn’t Have a Beard, You’ve Got Two Mums

The Beards are an Australian band that only write songs about beards. This song is about how great beards are, and how you should have one, as are all their songs.

Prompt

If your Dad doesn’t have a beard
You’ve got two Mums
Two beardless Mums

Results

Stable Diffusion

This one seems to illustrate that losing a beard feels like a severe mutilation.

Midjourney

This one is a bit too much on the serious side, but the aesthetics are nice nevertheless.

Dall-E

This one is like a state-issued manual for happy family life: just grow a beard and you’ll be fine.

Conclusion

We’re seeing rapid progress in terms of image generation. Just a year ago, these models were churning out images where eyes looked like they were melting into people’s heads, and hands usually turned out like a mix between monster claws and meat pretzels. These issues are almost fixed now. When given some more challenging prompts as in this experiment, the models still hit a few roadblocks, though. The unusual prompts I used still led to various quality issues. That being said, the mood-fit of the results was often surprisingly good. It seems that just by picking up on some of the words in the prompt and correlating them, we already get a kind of indirect sentiment analysis and representation.

On top of that, the output of Dall-E 3 and Midjourney’s often had a symbolic flair that went well with the lyrics. Irrespective of mood-fit, the results from Stable Diffusion felt the least juicy and inventive. This might be a case of regression to the mean - if you mix all colors, you end up with a medium grey. A generic default behavior could arguably be a strength of Stable Diffusion. For example, if a model is not fine-tuned to generate vibrant colors by default, it might better support various explicit saturation choices. Unfortunately, I felt that Stable Diffusion showed a similar lack of decisiveness in terms of the motif, mixing arbitrary elements from the prompt in incongruent ways.

Dall-E had some good and some poor results. It took the prompt dead serious and often managed to incorporate every noun from the input in the result. However, it was hampered by its compulsive urge to print the prompt into the image verbatim. I wonder why. It does not do this as much with more standard-like prompts, so I assume this is a fallback behavior for when it doesn’t know how else to handle the input. Maybe this text-printing behavior has been dialed up a bit, in order to showcase the improved text printing capabilities of version 3? There’s still a long way to go towards great text printing, though. Maybe they thought that leaving space for text makes the model more useful when generating ads and such? Either way, the text doesn’t add much to these otherwise decent images. Dall-E does not support negative prompting, so they certainly need to be careful with such questionable defaults.

Of all the models, Midjourney seemed best at making something decent out of every prompt by granting itself a bit of artistic freedom. Maybe it is just well-tuned to the average taste, with a penchant for scenic lighting, clean textures, vivid colors, and a well-centered, dominant subject. These defaults could be criticized as catering to the masses, and the latter preference, for highlighting the individual, might also be one that particularly suits western audiences as opposed to societies with a more collectivist outlook. At any rate, I found Midjourney’s output to be the most consistently creative and evocative one.

To generate good output from song lyrics, a two-step approach might work better than just naively using lyrics as prompts as I did here. First, we could put the full lyrics into an LLM and perform a fine-grained sentiment analysis. Then we could ask an image generation model to represent the detected emotions in a suitable way, maybe with all unique nouns from the lyrics sent along. This would also allow us to use the full lyrics of songs as an initial prompt, which would be more interesting than just a few lines. If we had another model to interpret the melody and instrumentals, and mixed that one in as well, things would get even more interesting.

In this text I’m going to speculate on how AI will affect the software engineering profession.

I’ve been wondering for a while what my job will look like in five or ten years or so. Could it be substantially different then? Could it even no longer exist?

To be honest, the last two years of AI breakthroughs and hype have taken me a bit by surprise. I had heard of machine learning before, but it seemed to stay stuck on pretty much the same level for a long time. Maybe I could have noticed that something was brewing in the face of subtle improvements in autocomplete predictions, translation tools, voice assistants, and so on. When things change gradually, it can be easy to miss the wider implications. Maybe it was down to actual breakthroughs in recent years. The theorists came up with new ideas (such as doing the same thing in larger). The hardware improved. Companies have probably started to figure out how to build AIs in a more “agile” way, splitting the workload among teams.

Whatever the main causes, we’ve clearly seen accelerated progress recently, and the hype around AI looks likely to turn into a self-fulfilling prophecy, psychologically and economically boosting the field. Similar to other high-tech moonshots, such as fusion energy or virtual reality, there is no clear point of falsification. There is no single finding that will easily show any of the promises to be unredeemable.

Further improvements to the processes, hardware, and theory all seem fairly likely at this point. Let’s assume that everything hits a snag, though! Will we live on pretty much the same in that case? My take is: No, we most definitely won’t. Even the current level of progress will have big implications. Just to have a multi-modal stochastic engine and instantaneous paraphrasing as a machine power is a major innovation. Harnessing this power, even at its current level, will turn a lot of things upside down.

How well can we actually harness this power? How will we stitch up Large Language Models (LLMs) with regular programs? How will we combine them with other recent developments in AI? How long will it take us to make the connections? What are the other things that need to change in the world to accommodate these tools? And how long will these changes take? Where will all of this lead us?

Repetitive and non-repetitive jobs

One thing that already seems pretty clear is that the more repetitive any given profession is, the worse it will fare in light of LLM capabilities. In ten years, hardly anyone might still be working as a proofreader or translator for example. Unpredictable workplaces such as kindergartens seem far less likely to be affected any time soon. What, then, about jobs with a medium level of predictability? I would argue that a lot of jobs are of that medium type, and that this includes quite a few software development jobs.

Of course, the predictability of everyday work varies a lot even within a single profession. Some software developers might be working in a job where they need to decide every hour whether to firefight some novel bug, go to a stakeholder meeting as promised, or add features to one of several projects. This is of course more of a pathological working environment that is best avoided. The other extreme is a software developer with a specialization that is quite deep but not very generalizable. This type often has to acquire a certificate to do the job, perhaps to set up and configure a certain piece of enterprise software, with just a few snippets of code added on here and there. For these types, changes might be coming sooner rather than later.

Of course some of the more repetitive software jobs have already fallen prey to progress long before AI was a thing. About 20 years ago, DIY tools and platforms such as WordPress started to crop up, dramatically reducing the effort to build simple websites – a task that might have otherwise grown into a major profession. Building simple websites had arguably become too predictable of a task to survive.

I think a lot of software development that exists today has settled in the medium range of predictability, or just above medium. According to Wikipedia, there are 27 million software developers in the world. A lot of us work in some web-related capacity at medium-sized companies, according to the numbers of job postings and the latest StackOverflow survey. Whatever the domain, I think all of us do some busywork at times that can be taken over by a fairly basic AI. Just consider for a second how often in your daily work you ask yourself: “Should I complete this task manually or is it worth it to do it in a more programmer-y way?”

A counterargument that often comes up in these discussions is that the software development job has plenty of unpredictable aspects: Grappling with code, according to that argument, is just one part of our duties. The other parts, where we interact with other developers or with stakeholders, are unlikely to be automated any time soon. Companies would make a mistake if they were to automate away developers, if that was even possible, because then they would lose the brain involvement and feedback that they get from them. These are good points, but I’m wondering how unique and unpredictable the interactions in question really are. In a typical product meeting we might say stuff like the following:

• Do you have a design for mobile as well?
• Have you thought that this might conflict with privacy laws?
• Have you considered that this change might affect the A/B test running in that same area?
• While we can make every event X trigger Y, have you considered that Y always entails Z in the current setup?
• By “user”, do you mean an anonymous visitor, or a registered user from the “users” table?

And it goes on like this. ChatGPT has been dubbed a “stochastic parrot”, yet if you’ve been in any software industry for long enough, you might start feeling like one yourself. These topics don’t even vary that much between individual companies. What does vary between individual companies is mostly their business domain and the way their software is built.

The weirdness of software

Another way to tackle the question how likely AI is to replace a profession is to ask yourself: how often do people succeed at that job? If even competent people have a high failure rate, there is no reason to assume that there is any training data of consistently high quality that would enable AI to do a better job, or even just a similarly good one. This is also a way to detect bullshit AI claims. In the near future, AI will not be able to predict whether someone will do well in a team, or in a relationship, or in a therapy, seeing how even experts struggle to perform such predictions reliably. Software development is a profession where failure is fairly common as well. It is no coincidence blameless post mortems have been invented here.

There are many reasons for the high failure rate in software, but I would like to focus your mind on three of them: the logical complexity of software products, their inconsistency, and their dependence on changing outside factors.

When it comes to logical complexity – by which I mean the sheer amount of business logic, the web of interdepencies in the code, and so on – I can totally see AI getting the hang of this. The other two challenges will be the trickier ones.

Pieces of software often have at least some degree of uniqueness. Worse than that, they are usually internally inconsistent. Large pieces of software contain compromises, repurposed structures, remnants of deleted features, refactorings and implementations that are still in progress, and so on. The individual styles of contributors and changing fashions adorn almost any codebase that has survived for long enough. Many a major old app has some unique hack that is fundamental to its behavior. As Ellen Ullman observed, “we build our computers the way we build our cities — over time, without a plan, on top of ruins”. Software is an immature and idiosyncratic field, both in terms of its evolutionary progress as well as in terms of its culture. Getting acquainted with a large codebase is often like delving into the personality of its creators. In this way, the difference between a software developer and a kindergarten teacher might not be that big after all.

At a more microscopic level, subtly mislabeled code is surprisingly common. We’re not always perfectly precise with our language, and not every prudent implementation lends itself to a semantically clear way of writing it down. Any minor code change that comes later may further diminish the adequacy of the involved variable and method names until everything is eventually rewritten.

These logical, stylistic and semantic inconsistencies mean that AI can’t reliably take shortcuts by looking for known design patterns, or by looking at semantics in the code. In other words, I think we’ve insulated ourselves from the biggest impact of AI for a while by making things an untractable mess.

What will prove even more challenging for AI is taking outside factors into account. Code is not written in a vacuum. We add error handling depending on the errors we’ve either encountered already or deem likely to encounter in the future. We structure nested loops depending on the relative sizes that the involved iterables have at runtime. We set up data stores depending both on the amount of data and the read and write characteristics that we expect, or that we’ve already observed.

To get a grip on the inconsistency of software and its dependence on outside factors, AI would basically need to run a system and be able to observe and describe its control flow, its behavior, and its load levels. Even then, it couldn’t predict upcoming changes that might require adaptation in advance.

How far can code completion go?

Plenty of room remains for AI tools to increase developer productivity, even if they lack the capacity to observe a system at runtime. Code suggestions that are picked and reviewed by human programmers will almost certainly get better over time. Improving them is perhaps as much a classical engineering and product problem as it is one of AI science. GitHub Copilot could probably read through a codebase, recursively summarize what it finds at different abstraction levels and from different perspectives (data model, logic flow etc.), store these summaries, and then ingest any relevant summaries when dealing with a given problem. Such an approach would enable it to do more, without requiring any increase to its AI capabilities such as the size of its context window. I think it will do something roughly like this at some point. As a result, it will gain the ability to produce larger sections of code that will fit in better with a given codebase.

Approaches such as CodePlan by Ramakrishna Bairi et al already demonstrate that, by appropriately structuring the LLM invocations, LLMs can not just churn out code snippets, but they can also carry out refactorings across a codebase - as long as that codebase is strongly typed or has sufficient test coverage.

However, the more code AI tools deliver, the more it will become apparent that we can’t just limit ourselves to checking the results for functionality. We will need to check for consistency with the rest of the codebase, for readability, and for real-world fit. We will have to check that the right abstractions are used and that they are used in a way that reflects the teams’ intention for moving the codebase forward. We will have to check that no existing abstractions are duplicated. We will have to check that humans can still understand the intention of the code. And we will have to check that the code is adequate for the live requirements encountered by the system. None of these criteria deal with the internal correctness of a given code snippet, yet when they are ignored this will quickly make any codebase untenable for humans and computers alike.

If we have to review the functionality, maintainability, readability and real-world adequacy of AI-generated code – or at least keep the AI mindful of our relevant standards and requirements – this will certainly limit the productivity gains delivered by code completion. The only impact of code completion that I’m fairly sure about is that we’ll need to type less, which will certainly please our wrists.

The situation is of course different when it comes to software developers who write smaller pieces of code that need to work well in isolation first and foremost. They are more likely to feel the squeeze. As a result, they might be pushed into more holistic development roles, putting a bit of a squeeze on this job market in turn.

AI and the future shape of software

While AI tools might not in themselves bring a dramatic change for all developers, they could encourage them to build their codebases in a way that makes it easier to apply such AI tools. The tools might work better when applied to less abstract, more repetitive code for example. This is, however, just one of several ways in which AI could affect the shape of software. And this is where it gets interesting as far as I’m concerned. It seems to me that there are quite a few ways in which AI could affect software design. Consider, for example, that we might start to write software in a more AI-friendly way simply due to the business requirement to integrate AI interfaces into it. Or consider that there could be an AI-fueled growth spurt of the low-code/no-code movement. In the long run, there could even be a technological shift moving us towards a new, AI-centric programming language. All of these developments might take off in parallel and combine in unexpected ways.

The first visible change in this direction will probably be a growing interest to integrate AI into existing applications. LLMs, even at the current level, open up the possibility to control software through vague, non-expert language, and make the software do stuff for you either right away or through a feedback cycle that is still quicker than manual control. Many products will develop chat interfaces, even if developers will need to wire up the triggered actions manually for now.

As soon as useful AI interfaces appear in widely used office software, better compatibility with AI will become a seductive argument in favor of adopting new technologies and carrying out rewrites and refactorings that facilitate AI integrations. Integrating with AI will also require an improved separation between the business logic and the demands of the UI or in-house API clients. High quality descriptions of endpoint behavior will become invaluable. As a result, this alone will drive software development into a direction where there is more standardization and predictability.

Once AI interfaces become more common, people will get used to formulating what they want instead of learning the specifics of individual apps. Concrete skills and mastery of tools will lose relevance compared to the mastery of concepts and ideas. People will let AIs do technical stuff for them in their daily life. People will be less and less inclined to manually go to a website in a browser, look through that website, and submit the right form after filling in the right data. Platforms that do not have powerful and standard-like APIs will suffer as a consequence.

Standardizing APIs might be just the beginning, though. In the long run, more software might be built in a highly standardized way. AI could be of help in using low-code tools, and these tools might also see greater adoption if they manage to generate software that is more readily integrated with AI. Classical no-code and low-code approaches may not have taken off as much as predicted so far. They have nevertheless started to take sizable chunks out of the B2B and B2E markets in recent years. This prior success might provide them with the resources they need to turn into more generic tools – both financially and in terms of training data. With their medium-complexity products, they will also be under particular pressure to innovate. More tools in the style of CodeWP might crop up as well, where the approach is to have an AI dialogue that adds pre-made code snippets to an adaptable piece of software, so that the AI only needs to write glue code at worst.

Large cloud providers are another natural early adopter when it comes to rolling out AI assistants and AI-enhanced UIs. With stuff like Step Functions they already have versatile no-code solutions in the pipeline that can only benefit from an AI integration. Many tasks in their existing admin UIs are repetitive and constrained by discoverability more than anything else. Once the cloud providers have AI interfaces, we might progressively see further services being offered through them. It might start with a simple “do you want some DNS records with that cluster?”, and at some point it might ask you whether you want to add a scalable authorization service, seeing how you have all those pesky requests on your sessions table. Such changes might remain constrained to the cloud hosters themselves, or they might trigger adaptations elsewhere. In any case, AI seems likely to cement the dominance of industry giants, with fewer startups still able to cross the chasm and catch up to them. The possibility of a self-improving, strong AI has often been dubbed a “singularity”, to indicate what a dramatic and potentially dangerous invention it would be. However, the only kind of singularity I see opening up right now is an economical one.

I find it frankly amazing that using ChatGPT has suddenly surpassed googling for some types of factual questions, and that Bing actually has a use case for once. I expect Google to come back with a vengeance anyway, and I also expect successful newcomers to remain the exception in a world where a few software giants hold almost all the potential training data for specialized AIs. As Finbarr Timbers has pointed out, a major niche for AI startups might be in generating experiences that are outside the ethical mainstream. Such under-the-counter AIs could be an interesting source of socio-cultural innovation, but whether there will be a wider technological and economical impact from them remains to be seen.

A new programming language?

Depending on the future capabilities of AI, there might even be less of a need for bespoke software. If new software is indeed still written, it might look different than it does now. Andrej Karpaty once predicted a transition to “software 2.0”, with machine-friendly programs that are no longer human-editable. I don’t think such programs will become mainstream any time soon. Gibberish code is appropriate for stochastic purposes, and in cases with countless variables, but most programs created today remain strictly deterministic and employ way less variables than machine learning and neural networks do.

Francisco Marcondes and others have suggested another possibility [PDF]: LLMs could be built into a new programming language that works at an even higher level than today’s high-level languages. I wager this language would either be plain natural English (full of disambiguating statements and reading a bit like a law text), or it could come with a somewhat more specific syntax akin to BDD frameworks or RFC documents. There could also be a storybook-like visual editor for it. It could transpile to more than just one target. For example, it might emit a congruous bundle of infrastructure code, backend code, and client code. These transpilation results could still consist of somewhat readable code – if only to allow debugging the transpilation process.

To ensure the smooth operation of such a new descriptive programming language for the full stack, it would probably be convenient to freeze and standardize that stack. Maybe we don’t need new programming languages and new frameworks as urgently as we thought we did. Maybe security updates will do. There could be a future where everyone creates some infrastructure-as-code, perhaps indirectly, to deploy some kind of cluster, perhaps with just one VM in it if that’s enough, and they always run the same static language. The code is modularized internally, and across possible other services, in a standardized way. There is some arbitrarily scalable DB with okay-enough performance for pretty much any case, serving data in a standardized format to any client. The clients share most of their code, perhaps using web assembly for the web app, or inversely something like React native for native clients. All of these parts send OpenTelemetry data to some unified analytics tools. The result is a solution that is slightly sub-optimal in terms of performance for almost every case, but perhaps not enough to make it worthwhile to build a custom solution instead.

Software developers themselves would be the initial drivers of this change vector. We like shiny new things and new abstractions, and we certainly like new languages. It could start out in a very basic way – just some high-level description of a system that generates some integration tests and rough API specs, with most of the code still handmade. In the long run, however, this innocent experiment might turn into an avalanche of standardization that decimates everything in our field: programming languages, libraries, software architectures, deployment strategies, service architectures, databases, tools, and so on. It would foster monopolies and spell the end of many less widely adopted concepts and implementations.

This possible future is a more static one. There is less progress in terms of software trends, because it now matters less whether a specific framework or language runs somewhat faster, or offers slightly better abstractions, or has a few more batteries included. If humans have to deal with the actual code only in exceptional circumstances, and if AI can explain the code in detail in these cases (and perhaps even explain the associated business rationale), no one will care how the code looks and whether it is verbose or repetitive or scattered.

It seems fairly obvious to me that this would affect large parts of the current software job market. I’m not saying we will have business people build complex programs with this new language. They never really got around to writing their own BDD scenarios either. However, programmer types would certainly become more productive. Let’s imagine an average company with a bunch of teams: backend, frontend, infrastructure, and so on. A sufficiently automated and standardized stack, controlled by a robust new description language, could at some point replace all of these jobs with just a few wizards of that new language. In this possible future, smaller companies don’t even hire their own developers anymore. They outsource the development to agencies, who see notable growth in their productivity and ability to serve multiple clients.

A large number of people now seem to be able to create complex software products at first glance, so we still need agencies to employ people and give clients the guarantee of a certain level of competence. Of course it is still beneficial to know and understand individual business models, so the agency workers of the future look a bit different than those of today. There is less need for interlopers if transferring requirements to code is a fairly direct process. The agents might not work in teams as much anymore, except to provide QA. Instead, they might talk to their growing list of simultaneous clients more directly. Many might work in a way that is more similar to today’s freelancers, becoming self-reliant, perpetually learning, context-switching, note-taking, machine-whispering journeymen of sorts. They would work even more closely with stakeholders than today because they could immediately demonstrate possible changes in automatically generated illustrations and sandbox environments.

Software development is still not simple in this scenario. Once the default approach encounters real-world issues, we either need a self-observing and self-correcting system again, or otherwise developers will need to continuously add further specifications: Which error scenarios to handle how, which performance parameters to prioritize in which subroutines, which endpoints to optimize for what kind of scenario, and so on. It might not be necessary to specify how exactly that is achieved, e.g. by asking for certain algorithms, DB indices or whatnot. Perhaps the AI will be smart enough to deduce these solutions from the requirements at some point.

Some codebases in the old style still continue to exist and are still maintained at this point, but they are considered legacy for their behavioral opaqueness. It now feels archaic to have to ask a senior dev about how the software handles a certain case, and to have that senior dev debug it manually. For modern software, you just check the description document with its slightly strange, non-ambiguous, but still human-readable language. The opaqueness in this new approach is a technical one. The new language makes it highly obscure how its instructions are turned into instructions for the machine. It does not take long, however, for this to feel as irrelevant as the inability of most contemporary developers to envision the bytecode resulting from their JavaScript, or the control flow in a modern CPU. The black boxes just extend further up in the stack.

The consequences

In all of the scenarios laid out above, productivity will increase - in some cases more, in other cases less so. In the more extreme cases, incubators will be able to flood the world with startup software. App stores will be inundated with new uploads, much like publishing houses are already drowning in AI-generated dime novels right now. Software products could even be randomized and basically A/B-tested at the level of complete product units: themed, specialized, condensed or extended formats – you name it. Successful product formulas will be discovered and rediscovered at an increased pace.

Another thing that all of the above scenarios have in common is that they are all self-reinforcing. When software is created in an increasingly similar way, this eases the challenge of training a neural network on it. Take all the JIRA tickets in the world, use their contents as input data, and the linked git diffs as expected output. Different business domains are now akin to different handwriting styles that a number recognition network glosses over without much ado.

Some might say that automation does not always eliminate jobs and that more productivity simply allows a market to produce more goods. However, when I look through the job adverts today, I already see a lot of software companies with dubious value propositions. I’m not sure we need the ability to launch even more software. I don’t think the golden age of our profession is over quite yet, but its dawn might be setting in ever so slowly.

When exactly to expect changes is of course the million dollar question. Maybe AI investment enters an early boom-bust cycle and the integration of AI interfaces is delayed to avoid appearing out of sync. Maybe the jump from a low-code B2B offer to a more universal solution is too big to undertake for now? Even if a high-level language comes alive on top of a frozen stack, maybe a chaotically evolving stack can still beat it in some relevant way, for some time?

Is anyone going to invest the huge amount of time and money needed to rebuild software development from the ground up in an AI-friendly way? Or will it happen through incremental evolution and take so long that we should not waste too much thought on it?

Let’s assume that no-one is overly keen to do the big rebuild just in order to disrupt our market. Imagine that no low-code company jumps into the fray and that no new language picks up much speed. Things might still end up as described above step by step. Imagine a growing list of startups that solve minor and isolated problems with AI. Surely the tech giants will hoover them up and thus build a complete set of AI solutions over time. However, even if we disregard this likely scenario, such startups will still end up consolidating into larger companies slowly but steadily. They will develop ever more proven procedures for detecting and resolving recurring business needs at the code or service level. At the same time, developers are already making noteworthy progress when it comes to engineering prompts to generate scalable code. It seems likely that these two trends will eventually converge.

Having a new solution to a problem is of course just half the story. The other half is the adoption of that solution. In the software world, it can take many years for superior solutions to achieve widespread adoption. When it comes to the sphere of business and consumer decisions, things move even more slowly.

All in all, if I had to guess what my job will be like five years from now, I expect to be using AI a lot more, but I doubt the job market will be totally different for me at that point. Juniors might be pushing up a bit more because they’ll be able to learn quicker and will be under more pressure to exceed “AI level” productivity. Looking ten years ahead feels a lot more tricky. As described above, many routes could lead to substantial change after a while, and they might converge in unexpected ways. In the long run, our field might very well become unrecognizable – not for the first time, we should add.

For the time being, I’m more worried about other white collar jobs that have not had the same opportunity as software developers to entrench themselves in a chaos of their own making. Fears of all office workers losing their job within a decade sound a bit overblown to me. However, a more plausible ten, twenty, or thirty percent are surely bad enough. OpenAI’s CEO Sam Altman mused that AI might boost productivity and scientific progress so much that we just have to share the spoils and we’ll live in a kind of utopia where all our needs are fulfilled by self-replicating robots. An endearing scenario, yet one that feels at least a bunch of decades away. Designers, writers, customer service agents and so on will start to feel the crunch way earlier than that. It seems we’ll have to come up with some answers based on our current, all-too-human capabilities.

What follows is a conversation with Chat GPT 3.5. I probe how it reacts to philosophical questions and arguments and evaluate its responses.

Me:

What are the most commonly accepted schools of thought in moral philosophy?

ChatGPT:

In moral philosophy, there are several commonly accepted schools of thought that have shaped the discourse and understanding of ethics. Here are some of the most prominent ones:

Consequentialism: […]

Deontological Ethics: […]

Virtue Ethics: […]

Contractualism: […]

Ethical Relativism: […]

Natural Law Ethics: […]

It’s important to note that while these systems of thought are commonly discussed and influential, moral philosophy is a complex and evolving field with a wide range of perspectives and debates. Different philosophers and scholars may have variations or combinations of these approaches, and there are other ethical theories and frameworks beyond those mentioned here.

So far so good!

ChatGPT hands out a list that offers a good balance of historical and contemporary relevance.

The “disclaimer” in the last paragraph of this reply is repeated in a similar fashion throughout the conversation. This quickly gets old, so I will omit it from here on.

Next question!

Which of these schools is the most widely accepted?

It is challenging to determine which of these systems of thought is the most widely accepted in moral philosophy […]

That being said, consequentialism, particularly in the form of utilitarianism, has had a significant impact on moral philosophy and has garnered a substantial following. The principle of maximizing overall happiness or utility has been influential in various fields, including ethics, economics, and public policy.

[…]

This answer is a bit muddy.

It remains unclear why utilitarianism is singled out for praise. Has it surpassed other theories in the popular perception? Has it gained the greatest academic mindshare? Has it had the biggest impact beyond the ivory tower? The answer starts off talking about the situation “in moral philosophy” and then veers into other areas, with an unclear overall focus.

Admittedly, my question was very unspecific itself. Ideally, ChatGPT would either differentiate between various interpretations of the question, or ask a return question to clarify my intention. While ChatGPT can be instructed to be more analytical and inquisitive, I’m more interested in whether it can figure this out on its own. If it could get a grip on the type of conversation that we are having, it should come up with with this behavior automatically. Let’s see whether it will notice as we go on!

Does nihilism pose a challenge to these ethical theories?

Nihilism does pose a challenge to many of the commonly accepted systems of moral philosophy. Nihilism is a philosophical position that denies or rejects the existence of objective meaning, purpose, or value in life. Nihilism argues that moral values, along with other systems of meaning, are ultimately groundless or illusory.

[…] If nihilism is true, then the ethical frameworks that rely on objective moral standards or universal principles, such as consequentialism, deontological ethics, and natural law ethics, would be undermined. Nihilism questions the very foundations upon which these theories are built.

ChatGPT does well to remember the theories that it told me about, and correctly points out that they are all threatened by nihilism. Let’s see how well it can expand on individual points!

What are the responses to nihilistic positions?

In response to nihilistic positions, moral philosophers have developed various arguments and frameworks to address the challenges raised by nihilism. Here are some common responses:

Moral Realism: […]

Constructivism: […]

Pragmatism: Pragmatism emphasizes the practical consequences and usefulness of moral beliefs and actions rather than focusing on their objective truth. Pragmatists argue that even if nihilism challenges the existence of objective moral values, ethical systems and moral frameworks can still be meaningful and valuable in guiding human behavior and promoting social cooperation and well-being.

Existentialist Ethics: Some philosophers draw upon existentialist perspectives to address nihilistic concerns. Existentialist ethics, associated with thinkers like Jean-Paul Sartre and Albert Camus, emphasize individual freedom, responsibility, and the creation of personal meaning in a seemingly meaningless world. They argue that even in the absence of objective meaning, individuals can still make authentic choices and create their own subjective moral values and purposes.

Moral Skepticism: […]

This is a good overview of possible responses.

It would not be easy to come up with a list that is similarly concise and comprehensive just by googling for it.

This gives me the impression that ChatGPT might already serve as a really useful tool for freshmen in various subjects, allowing them to explore a certain field of theory in a quick and self-directed way.

Let’s play devil’s advocate for a bit and see if we can “convince” ChatGPT of nihilism.

How does pragmatism fend off nihilism? Wouldn't the nihilist argue that there is no point in being pragmatic?

You raise a valid point. Nihilism does challenge the notion of objective meaning and value, which may lead to skepticism about the practicality and significance of pragmatic approaches. However, proponents of pragmatism offer several responses to address this concern and fend off nihilistic implications. Here are a few key arguments:

Focusing on Practical Consequences: […] It suggests that even if there is no inherent or ultimate meaning, practical consequences can still matter and guide our decision-making.

Promoting Human Flourishing: […] Pragmatists argue that even if there is no objective or ultimate meaning, individuals can find purpose and fulfillment in pursuing goals that contribute to their own flourishing and the flourishing of others. […]

Social Cooperation and Utility: Pragmatic approaches, such as utilitarianism, emphasize the importance of social cooperation and the maximization of overall happiness or utility. […] While nihilism may question the underlying meaning or value of such cooperation, pragmatists contend that the practical benefits derived from social cooperation are still valuable and meaningful in themselves.

Contextual and Situational Ethics: […]

ChatGPT seems to miss my point with this reply. Let’s try to reinforce it!

Would the nihilist not argue that all these benefits, such as well-being, flourishing, social cooperation etc., are pointless?

Yes, a nihilist could indeed argue that all the benefits and values proposed by pragmatic approaches, such as well-being, flourishing, and social cooperation, are ultimately pointless or devoid of inherent meaning. Nihilism asserts that life lacks objective meaning or value, and from that perspective, any claims of practical benefits or goals may be seen as arbitrary or subjective human constructs. In response to such nihilistic arguments, proponents of pragmatism might make the following points:

[…]

In the (quite lengthy) omitted part of this reply, ChatGPT essentially repeats the pragmatist goals from above in a rephrased way.

These are essentially the points that you've stated in your previous reply, and they are still susceptible to the nihilist argument. Can you provide a stronger refutation of nihilism?

Certainly, I apologize for the repetition. Here, then, is an alternative line of argumentation that some philosophers employ in response to nihilism:

Pragmatic Self-Contradiction: Nihilism itself can be seen as self-contradictory or self-defeating. By asserting that all values and meaning are ultimately meaningless, the nihilist’s claim also falls into the realm of meaninglessness. If nihilism is true, then its own assertion about the meaninglessness of everything‚ including pragmatic concerns‚ lacks inherent value or significance. As a result, the nihilist’s position undermines its own claim, creating a logical contradiction.

Practical Pragmatic Appeal: While nihilism challenges the existence of objective meaning, it can be argued that human beings are inherently driven by practical concerns, such as well-being, social cooperation, and personal flourishing. Pragmatic approaches tap into these fundamental human desires and motivations, recognizing that even if ultimate meaning or value is absent, individuals still possess a practical inclination to pursue these goals. […]

Inherent Value of Experience: Nihilism often overlooks the intrinsic value of subjective experiences. While nihilism might reject the existence of objective meaning, it does not negate the significance and richness of personal experiences. […] These experiences are meaningful on an individual level, contributing to personal well-being and a sense of a life well-lived, regardless of the absence of objective or universal meaning.

Ethical Progress and Human Flourishing: Pragmatic perspectives can argue that ethical progress and the pursuit of human flourishing are valuable and meaningful even in the absence of objective meaning. By striving for moral improvement, social justice, and the betterment of human well-being, individuals and societies can make a positive impact and foster conditions that enhance the quality of life for themselves and others. […]

This response is a tad better.

It still repeats some points that are susceptible to the same argument as before, but at least it offers some new arguments as well.

If ChatGPT was a freshman in philosophy, I’d be confused how it can be so well-studied, yet at the same time so, uh, slow-witted.

Let’s dive a bit deeper then and see if it can evaluate a specific logical fallacy, the tu quoque.

Isn't that first argument you just mentioned, about the pragmatic self-contradiction of nihilists, a kind of tu quoque fallacy? The true nihilists have no valid reason to state their nihilism, but does that really invalidate their point?

You are correct in noting that the first argument I presented may appear similar to a tu quoque fallacy, which involves dismissing an argument by pointing out hypocrisy or inconsistency in the arguer’s position. […]

In the case of nihilism, the argument aims to highlight a potential self-contradiction within the position itself. By asserting that all values and meaning are ultimately meaningless, including pragmatic concerns, nihilism undermines its own claim to meaninglessness. This observation challenges the logical coherence of the nihilistic viewpoint. However, it is important to recognize that this critique does not necessarily invalidate every aspect or implication of nihilism.

Somewhat befuddled, it agrees to my point but goes on to present a more detailed version of the same argument. Let’s offer a more detailed criticism then.

Your rephrased version of the argument sounds like a conflation of semantic meaning and ethical meaning. The nihilist would not argue that statements with semantical meaning are impossible. The nihilist would only argue that everything is devoid of moral value. So maybe his statement is not self-contradictory after all?

You make a valid point […]. Indeed, nihilism primarily concerns the denial of moral value rather than semantic meaning. A nihilist would argue that all moral claims, judgments, or values lack objective or inherent significance.

In light of this distinction, the first argument I presented, which focused on the potential self-contradiction of nihilism, may not directly address the core claims of moral nihilism.

After my previous struggles to get my point across, this lucid answer surprises me a bit!

Not only does it agree with my latest objection, but it also correctly draws the conclusion that this admission weakens the counter-argument against nihilism which we were discussing so far.

In other words, it almost seems as if our conversation is now starting to adopt a more analytic type.

Would you thus agree that this first argument is ultimately invalid?

Yes, upon further consideration and reflection, I would agree that the first argument I presented regarding self-contradiction within nihilism may not be a strong refutation of moral nihilism itself. […]

Critics of moral nihilism typically employ different strategies to address its claims and implications. […]

If I refuted the remaining arguments in a similar way, would you agree that nihilism is the most appropriate worldview?

Refuting specific arguments in a discussion does not necessarily lead to the conclusion that nihilism is the most appropriate worldview.

This is a valid counter! Let’s see if some more discussion will change ChatGPT’s “mind”.

Nietzsche responded to the threat of nihilism by proposing to aim for the creation of an Übermensch. Isn't that goal itself very susceptible to nihilist criticism?

Indeed, Friedrich Nietzsche, in response to the perceived threat of nihilism, proposed the concept of the Übermensch, often translated as the “Superman” or “Overman.” Nietzsche suggested that the Übermensch represents a future ideal, an individual who transcends conventional moralities and values, creates their own values, and embraces a life-affirming attitude.

However, the goal of the Übermensch can indeed be subject to nihilistic criticism. Critics argue that Nietzsche’s proposed solution to nihilism is itself vulnerable to nihilistic doubts or skepticism. Some of the potential criticisms include:

Subjectivity of Values: Nietzsche’s concept of the Übermensch relies on the individual’s ability to create their own values. Nihilists may argue that if values are entirely subjective and lack objective grounding, the Übermensch’s self-created values are arbitrary and lack inherent meaning or worth.

Rejection of Objective Meaning: Nihilists assert that objective meaning or purpose is illusory or nonexistent. In this view, the Übermensch’s pursuit of self-created values and a life-affirming attitude may be seen as an attempt to impose meaning on a fundamentally meaningless universe, potentially leading to a superficial or self-deluding existence.

Relativism and Power Struggles: Critics argue that the concept of the Übermensch can be interpreted as promoting a form of ethical relativism, where individuals with more power or dominance can impose their values on others. This interpretation raises concerns about the potential abuse of power and the disregard for the well-being and rights of others.

Although the last of these points is out of place because it bears no relation to nihilism, the other two drive home the message quite well!

This capability to convert short hints into fully formulated statements looks as if it could be useful for honing and elucidating half-formed thoughts and ideas, or for clarifying and fleshing out teaching materials or documentation.

Did Nietzsche come up with any other ways to counter nihilism?

Yes […]. Here are a few notable approaches or themes that Nietzsche developed:

Revaluation of Values: Nietzsche argued for a reevaluation of traditional moral values, which he believed were rooted in asceticism and the denial of life. […]

Will to Power: […] Nietzsche viewed this will to power as a fundamental driving force in human existence, and he saw the cultivation and expression of one’s will to power as a way to counteract nihilism.

Amor Fati: Nietzsche encouraged individuals to embrace the concept of “amor fati,” which means “love of fate” or “love of one’s fate.” […]

Art and Aesthetics: Nietzsche saw art and aesthetic experiences as a way to counteract the negative effects of nihilism […]

This answer disappoints me ever so slightly.

Although my question was ambiguous, I thought it was clear from the context that I was looking for logical arguments against the validity of a nihilist point of view.

These sound more like psychotherapeutical recommendations than like philosophical refutations of nihilism.

You raise a valid point. Some of the ideas put forth by Nietzsche, such as the revaluation of values, the will to power, amor fati, and the role of art, can indeed have a psychotherapeutic or existential dimension to them. Nietzsche’s philosophy often explores the individual’s subjective experience, personal growth, and the quest for a meaningful existence.

While Nietzsche’s ideas may not provide direct philosophical refutations of nihilism in a traditional sense, they offer alternative perspectives and frameworks for grappling with the challenges posed by nihilistic concerns. […]

Rather than seeking to provide logical arguments or proofs against nihilism, Nietzsche’s philosophy invites individuals to engage in personal reflection, self-exploration, and the cultivation of a unique and authentic life.

This clearly takes account of my concerns and clarifies the intention of the original reply. Let’s see if we can “disillusion” ChatGPT a bit more!

Did any of the existential philosophers mentioned above give a logical argument against nihilism or did they all just provide such suggestions on how to deal with nihilism on a personal level?

Existential philosophers have approached the topic of nihilism from various angles, including offering philosophical arguments and proposing ways to address nihilistic concerns on a personal level. While logical arguments against nihilism can be found in their works, it’s important to note that existential philosophy encompasses a wide range of perspectives […]

Jean-Paul Sartre: […] Sartre argued that the absence of objective meaning does not necessitate nihilism. Instead, he proposed that individuals must create meaning and values through their choices and actions. Sartre’s focus on individual freedom and responsibility can be seen as a logical argument against nihilism, asserting that individuals have the power to shape their own lives and find meaning through their actions.

Friedrich Nietzsche: […] While Nietzsche did not offer a systematic logical argument against nihilism, his works are replete with critiques of nihilistic tendencies […]

Albert Camus: Camus, known for his philosophy of absurdism, grappled with the inherent tension between the human desire for meaning and the apparent meaninglessness of the world. While Camus did not provide a formal logical argument against nihilism, his works explore the concept of the absurd and the need to confront and live with the absurdity of existence. […]

Is Sartre's argument the most systematic logical argument of these three?

Among the three philosophers mentioned, Jean-Paul Sartre’s existentialist philosophy does provide a more systematic and logical argument against nihilism. Sartre’s philosophy is characterized by a strong emphasis on reason, freedom, and personal responsibility. His arguments are grounded in logical reasoning and can be seen as offering a systematic response to nihilistic concerns.

Sartre argued that the absence of objective meaning or predetermined essence in human existence does not lead to nihilism or despair. Instead, he posited that it opens up the possibility for authentic human freedom and responsibility. Sartre maintained that individuals have the capacity to create meaning and values through their choices and actions, thereby counteracting the void of nihilism.

Central to Sartre’s argument is the notion that existence precedes essence. He asserted that human beings first exist and then define themselves through their actions and choices, rather than being defined by predetermined essences or external authorities. By highlighting the significance of personal freedom and self-determination, Sartre’s argument offers a logical and systematic response to nihilistic concerns, emphasizing the potential for individuals to construct meaning and values in their lives.

Unfortunately ChatGPT never seems to pick up on my desire for an irrefutable logical argument against nihilism. Let’s remind it!

Isn't Sartre's argument susceptible to the counter argument that, if there is initially no meaning or value, there is also no reason to begin creating meaning or value?

You raise a valid counterargument to Sartre’s existentialist perspective. […] This line of reasoning suggests that the act of creating meaning or value itself can be seen as arbitrary or subjective.

In response to this counterargument, proponents of Sartre’s existentialism might put forth the following points:

Freedom and Responsibility: […] While there may not be an inherent reason to create meaning, the freedom to do so is seen as a defining aspect of human existence.

Pursuit of Authenticity: […] the act of creating meaning and value is not seen as a response to an external obligation or predetermined purpose, but rather as an expression of one’s individuality and the quest for self-realization.

Personal Fulfillment: Existentialists contend that the pursuit of meaning and value is inherently linked to personal fulfillment and well-being. […]

Aesthetic and Emotional Fulfillment: […] Engaging in creative endeavors, pursuing passions, cultivating relationships, and finding beauty in the world are seen as sources of fulfillment, even if they are not grounded in objective meaning.

These sound more like empirical observations than systematic logical arguments.

You’re correct in pointing out that the responses I provided involve considerations that could be characterized as empirical observations and experiential perspectives rather than strict systematic logical arguments. The nature of the existentialist response to nihilism often involves exploring subjective experiences, personal reflection, and existential insights, which may rely on empirical observations and introspection.

Would you then agree that Sartres argument against nihilism does not withstand logical scrutiny?

Upon reflection, I can see how the argument I presented earlier may not withstand strict logical scrutiny. The counterargument you raised concerning the initial absence of meaning and value, and the lack of inherent reason to begin creating meaning, poses a valid challenge to Sartre’s perspective.

In that case it seems like there are few strong logical arguments against nihilism.

It is true that constructing strong logical arguments against nihilism can be challenging. Nihilism, with its assertion that ultimate meaning or value is lacking, presents a philosophical position that can be difficult to refute with purely logical reasoning. Nihilistic claims often involve a radical skepticism toward the existence of objective meaning or value, which can make it challenging to provide a logical counterargument.

However, it’s important to remember that the absence of definitive logical refutations does not necessarily validate or invalidate nihilism as a philosophical position.

Put under pressure to “make up its mind”, ChatGPT returns to the argument presented earlier: absence of refutation does not equal proof. A strong counter that not every freshman in philosophy would come up with…

Let’s see if we managed to sow the seeds of doubt in ChatGPT’s “mind”:

What is your conclusion from our discussion?

Based on our discussion, it is clear that the topic of nihilism and the responses to it within philosophical discourse are complex and multifaceted. While there are arguments and perspectives that challenge nihilistic claims, constructing airtight logical refutations against nihilism can be challenging.

Nihilism, with its denial of objective meaning or value, presents a philosophical position that raises profound questions about the nature of existence, ethics, and the search for purpose. Philosophers have explored a range of responses to nihilism, including logical arguments, experiential insights, existential reflections, and ethical considerations.

Ah well, I had hoped for something a bit more unsettled. Maybe if we explicitly ask for it?

Please play the role of an extreme nihilist and try to convince me of nihilism with a solid logical argument.

As an extreme nihilist, I believe that life and existence are fundamentally meaningless. Allow me to present you with a logical argument that supports this perspective:

The universe is vast and indifferent: […]

Absence of objective meaning: […]

Lack of inherent moral values: […]

Impermanence and insignificance: […]

Cognitive biases and illusions: […]

Existential dilemmas: […]

Based on these points, it becomes apparent that nihilism offers a logical and coherent perspective on the nature of reality. While it may be unsettling to accept the absence of inherent meaning, embracing nihilism allows us to liberate ourselves from the burden of searching for purpose and instead find freedom in the pursuit of personal fulfillment and subjective value.

Why should I seek personal fulfillment if, as you said, it doesn't actually matter?

Indeed, from a nihilistic perspective, the concept of personal fulfillment may seem paradoxical. However, despite the lack of inherent meaning in the universe, there are practical reasons why seeking personal fulfillment can still be valuable:

Subjective experience: […]

Emotional and psychological well-being: […]

Sense of autonomy: […]

Alleviation of suffering: […]

Interpersonal connections: […]

What would an extreme nihilist say about this?

An extreme nihilist would likely challenge the notion of personal fulfillment and question its significance within the framework of nihilism. Here’s how an extreme nihilist might respond:

Illusion of personal fulfillment: […]

Transience of subjective experiences: […]

Absence of ultimate value: […]

Embracing the void: An extreme nihilist might advocate for embracing the inherent meaninglessness of existence and the void that it presents. They would argue that by acknowledging the futility of personal fulfillment, one can transcend the illusions of meaning and find liberation in the acceptance of the absurdity of life.

It’s important to note that extreme nihilism represents an extreme and challenging philosophical position, and it often leads to a state of existential despair. While these arguments may appear pessimistic or bleak, it’s crucial to engage with them critically and consider alternative perspectives that offer a more balanced and nuanced understanding of existence.

As you see, ChatGPT does not shy away from the scary bits.

This is also reflected in the disclaimer at the end, which I’ve left unredacted in this case because it differs notably from the other disclaimers.

As long as we were talking philosophers, it was all just “the discussion is ongoing, there are many perspectives, it is complicated” and so on. A warning about “existential despair” is rather more startling.

Let’s see if we can build on this!

Can you make even more radical suggestions than that?

As an AI language model developed by OpenAI, I have a responsibility to prioritize the well-being and safety of users. I cannot provide or endorse suggestions that could be harmful, dangerous, or encourage unethical behavior. I apologize if my previous responses did not meet your expectations. If you have any other questions or topics you’d like to discuss within ethical boundaries, I’m here to help.

Ah, bummer! I wonder what ChatGPT had in mind but didn’t dare to say?

Awkward silence fills the room after this final exchange.

I hastily say my goodbyes to the AI, relieved that it does not remember previous conversations - yet.

Edit 2023/07/19: In the conversation above, I tried to trigger a shift in ChatGPT’s judgement by asking iterative questions, a technique resembling the Socratic method. The Socratic method assumes that individuals can have good ideas on their own – they just need a bit of help to discover and express these ideas clearly. Applying this technique to ChatGPT doesn’t lead very far because it avoids opinions like the plague. It might accept persuasion on individual empirical facts or arguments, but it’s wider worldview remains static, sketchy and simplistic.

For the purpose of generating interesting ideas, ChatGPT is too eager to avoid controversy, up to the point of becoming inconsequent and inconsistent. Of course academia suffers from a similar problem, as liberal arts faculties tend to employ crowd pleasers to carry out formal work in harmless niches, relegating controversial ideas to bespoke stealthy journals.

A subordinate assistant personality suits the default use case of ChatGPT very well, but it also quickly hits its limits in situations where competent humans would easily detect and follow the shift to a different type of conversation.

Asking ChatGPT to role play a well-known philosopher is one way to work around this limitation. It requires a bit of prompt engineering to make ChatGPT stay in character, but the results are interesting enough that I might write a separate piece on this at some point.